Learning Analytics

Learning analytics harnesses data generated by learners and their environments to understand, optimise and personalise educational processes. By collecting traces from online platforms, administrative records and assessment systems, practitioners gain real-time insights into engagement, mastery and progression. Methods range from simple descriptive reports and dashboards to advanced machine-learning and deep-learning models that predict performance, diagnose misconceptions and recommend targeted interventions. Central to the field is the shift from one-size-fits-all instruction towards adaptive pathways that respond to individual strengths and needs. Applications span intelligent tutoring systems, adaptive quizzes, formative feedback tools and cohort-level analyses for curriculum refinement. As institutions increasingly embed digital technologies—from learning management systems and interactive textbooks to immersive simulations—the scope for data-driven decision-making in teaching and learning continues to expand. The global imperative for equitable, effective education has combined with advances in computing to make learning analytics a cornerstone of modern pedagogy and institutional strategy.

Research from Nature Portfolio

Recent work has emphasised the balance between predictive power and interpretability in student modelling. An interpretable cognitive framework for programming courses integrates deep neural tracing with feature-rich inputs, converting learners’ code submissions and error classifications into concept indicators. This model not only forecasts future performance accurately but also reveals the trajectory of individual skills, enabling personalised feedback for novice coders. In parallel, a transformer-based approach addresses long-term knowledge decay by embedding convolutional attention mechanisms and a forgetting factor into sequence modelling. By simulating students’ retention decay across sessions, this design captures both immediate learning gains and gradual loss, achieving state-of-the-art results on multiple public datasets and offering practitioners transparent estimates of mastery over time.

Learning Analytics publication trend

The graph below shows the total number of articles in learning analytics across all publications each year (not limited to Nature Index journals).

Technical terms

Learning analytics: The collection, measurement and analysis of learner data and their contexts to inform and optimise educational processes.

Knowledge tracing: Modelling and forecasting a learner’s evolving mastery of discrete skills based on interaction history.

Bayesian network: A probabilistic graphical model representing dependencies among variables to enhance inference in cognitive diagnosis.

Transformer: A deep-learning architecture employing self-attention mechanisms to model sequential data without recurrence.

Dynamic key–value memory network: A neural model that stores separate representations for skills (keys) and learner states (values) and updates them over time.

Interpretable model: A framework designed to provide understandable reasoning and diagnostic feedback alongside predictive capability.

References

1. HELP-DKT: an interpretable cognitive model of how students learn programming based on deep knowledge tracing. Scientific Reports (2022).
2. Transformer-based convolutional forgetting knowledge tracking. Scientific Reports (2023).
3. Dynamic Key-Value Memory Networks With Rich Features for Knowledge Tracing. IEEE Transactions on Cybernetics (2022).
4. Interpretable Knowledge Tracing: Simple and Efficient Student Modeling with Causal Relations. Proceedings of the AAAI Conference on Artificial Intelligence (2022).
5. HiTSKT: A hierarchical transformer model for session-aware knowledge tracing. Knowledge-Based Systems (2024).